Type I error rates in association versus joint linkage/association tests in related individuals

Jack W. Kent, Thomas D. Dyer, Harald H H Göring, John Blangero

Research output: Contribution to journalArticle

12 Citations (Scopus)

Abstract

Positional gene discovery on pedigree data typically involves initial gross localization by linkage analysis with subsequent finer localization by association analysis in areas that show evidence of linkage. We examine the effect of including linkage information when testing for association in the context of variance-components-based pedigree analysis. We present simulation experiments showing that, at least in the extreme case of a rare private allele, failing to include the linkage variance component in the association model results in excessive Type I error that increases with allele copy number and/or quantitative trait locus (QTL) effect size. Joint estimation of the linkage variance component in the association model reduces Type I error to nominal expectations. This holds whether allele-sharing probabilities are estimated from a polymorphic marker or from the very single-nucleotide polymorphism (SNP) being tested for association, although the latter provides much less power. These results support the idea that an appropriate association analysis must test both the random effect of shared marker alleles (linkage) and the mean effects of the marker genotypes (association).

Original languageEnglish (US)
Pages (from-to)173-177
Number of pages5
JournalGenetic Epidemiology
Volume31
Issue number2
DOIs
StatePublished - Feb 2007
Externally publishedYes

Fingerprint

Joints
Alleles
Pedigree
Quantitative Trait Loci
Genetic Association Studies
Single Nucleotide Polymorphism
Genotype

ASJC Scopus subject areas

  • Genetics(clinical)
  • Epidemiology

Cite this

Type I error rates in association versus joint linkage/association tests in related individuals. / Kent, Jack W.; Dyer, Thomas D.; Göring, Harald H H; Blangero, John.

In: Genetic Epidemiology, Vol. 31, No. 2, 02.2007, p. 173-177.

Research output: Contribution to journalArticle

Kent, Jack W. ; Dyer, Thomas D. ; Göring, Harald H H ; Blangero, John. / Type I error rates in association versus joint linkage/association tests in related individuals. In: Genetic Epidemiology. 2007 ; Vol. 31, No. 2. pp. 173-177.
@article{05bb25e2082f4ad6856d91611bfc0aa8,
title = "Type I error rates in association versus joint linkage/association tests in related individuals",
abstract = "Positional gene discovery on pedigree data typically involves initial gross localization by linkage analysis with subsequent finer localization by association analysis in areas that show evidence of linkage. We examine the effect of including linkage information when testing for association in the context of variance-components-based pedigree analysis. We present simulation experiments showing that, at least in the extreme case of a rare private allele, failing to include the linkage variance component in the association model results in excessive Type I error that increases with allele copy number and/or quantitative trait locus (QTL) effect size. Joint estimation of the linkage variance component in the association model reduces Type I error to nominal expectations. This holds whether allele-sharing probabilities are estimated from a polymorphic marker or from the very single-nucleotide polymorphism (SNP) being tested for association, although the latter provides much less power. These results support the idea that an appropriate association analysis must test both the random effect of shared marker alleles (linkage) and the mean effects of the marker genotypes (association).",
author = "Kent, {Jack W.} and Dyer, {Thomas D.} and G{\"o}ring, {Harald H H} and John Blangero",
year = "2007",
month = "2",
doi = "10.1002/gepi.20200",
language = "English (US)",
volume = "31",
pages = "173--177",
journal = "Genetic Epidemiology",
issn = "0741-0395",
publisher = "Wiley-Liss Inc.",
number = "2",

}

TY - JOUR

T1 - Type I error rates in association versus joint linkage/association tests in related individuals

AU - Kent, Jack W.

AU - Dyer, Thomas D.

AU - Göring, Harald H H

AU - Blangero, John

PY - 2007/2

Y1 - 2007/2

N2 - Positional gene discovery on pedigree data typically involves initial gross localization by linkage analysis with subsequent finer localization by association analysis in areas that show evidence of linkage. We examine the effect of including linkage information when testing for association in the context of variance-components-based pedigree analysis. We present simulation experiments showing that, at least in the extreme case of a rare private allele, failing to include the linkage variance component in the association model results in excessive Type I error that increases with allele copy number and/or quantitative trait locus (QTL) effect size. Joint estimation of the linkage variance component in the association model reduces Type I error to nominal expectations. This holds whether allele-sharing probabilities are estimated from a polymorphic marker or from the very single-nucleotide polymorphism (SNP) being tested for association, although the latter provides much less power. These results support the idea that an appropriate association analysis must test both the random effect of shared marker alleles (linkage) and the mean effects of the marker genotypes (association).

AB - Positional gene discovery on pedigree data typically involves initial gross localization by linkage analysis with subsequent finer localization by association analysis in areas that show evidence of linkage. We examine the effect of including linkage information when testing for association in the context of variance-components-based pedigree analysis. We present simulation experiments showing that, at least in the extreme case of a rare private allele, failing to include the linkage variance component in the association model results in excessive Type I error that increases with allele copy number and/or quantitative trait locus (QTL) effect size. Joint estimation of the linkage variance component in the association model reduces Type I error to nominal expectations. This holds whether allele-sharing probabilities are estimated from a polymorphic marker or from the very single-nucleotide polymorphism (SNP) being tested for association, although the latter provides much less power. These results support the idea that an appropriate association analysis must test both the random effect of shared marker alleles (linkage) and the mean effects of the marker genotypes (association).

UR - http://www.scopus.com/inward/record.url?scp=33846418780&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=33846418780&partnerID=8YFLogxK

U2 - 10.1002/gepi.20200

DO - 10.1002/gepi.20200

M3 - Article

C2 - 17219375

AN - SCOPUS:33846418780

VL - 31

SP - 173

EP - 177

JO - Genetic Epidemiology

JF - Genetic Epidemiology

SN - 0741-0395

IS - 2

ER -